ARTICLE Received 13 Dec 2015 | Accepted 1 Dec 2016 | Published 20 Jan 2017
DOI: 10.1038/ncomms14138
OPEN
The shape of telephone cord blisters Yong Ni1, Senjiang Yu2, Hongyuan Jiang1 & Linghui He1
Formation of telephone cord blisters as a result of buckling delamination is widely observed in many compressed film-substrate systems. Here we report a universal morphological feature of such blisters characterized by their sequential sectional profiles exhibiting a butterfly shape using atomic force microscopy. Two kinds of buckle morphologies, light and heavy telephone cord blisters, are observed and differentiated by measurable geometrical parameters. Based on the Fo¨ppl-von Ka´rma´n plate theory, the observed three-dimensional features of the telephone cord blister are predicted by the proposed approximate analytical model and simulation. The latter further replicates growth and coalescence of the telephone cord into complex buckling delamination patterns observed in the experiment.
1 CAS
Key Laboratory of Mechanical Behavior and Design of Materials, Department of Modern Mechanics, University of Science and Technology of China, Hefei, Anhui 230026, China. 2 Department of Physics, China Jiliang University, Hangzhou 310018, China. Correspondence and requests for materials should be addressed to H.J. (email: [email protected]). NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
1
ARTICLE
D
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
etailed origins of postbuckling phenomena have been a subject of discussions for several decades1. The buckles usually involve multiple instabilities and exhibit intriguing morphologies2,3. When a film deposited on a substrate is subject to a large residual compression, it buckles and delaminates away from the substrate, and the blisters with straight-sided, circular, telephone cord (TC) or networklike patterns are formed4,5. Predicting the shape of such blisters is of scientific interest for stress-driven pattern formation and important for avoiding structural failure in various film-substrate systems4. Among various manifestations of this phenomenon, the TC blister is the most frequently observed in many different film-substrate systems5. This motivates a search to identify universal features of TC blisters. Recently, TC formation has been considered as a development from a secondary instability of straight-sided or circular blisters. The instability condition has been analytically addressed by the Fo¨ppl-von Ka´rma´n (FvK) plate model6–9. Further, it has been found that the TC blister can also grow with the appearance of sags by an oscillatory process and its edge unnecessarily starts straight10,11. So far, a convincing solution for the three-dimensional (3D) TC blister morphology remains a challenge because of limited methods for handling nonlinear elasticity of thin plate, although a series of numerical models and experiments have provided valuable insights into the morphological features of TC blisters10–25. The elegant instability analysis of a straight-sided blister predicted a constant undulation-period-to-width ratio of the TC buckle under the clamping boundary conditions close to the experimental observation8. In addition, the ratio may be not constant and change with the interfacial adhesion and the film-to-substrate modulus ratio11–16. The zigzag undulation feature of the TC buckle has been replicated by numerical investigations5,15–19. A comprehensive study for the 3D TC morphology suggested that it could be approximately viewed as a sequence of connected segments of a circular buckle pinned at its centre12. However, such approximation predicted the ridge of the TC blister to be discontinuous, different from our following refined experimental observations wherein the ridge is continuous and its height is periodically changing. Here we aim to probe the 3D features of the TC morphology using a combination of analytical model, numerical simulation and atomic force microscopy (AFM) study. As shown in Fig. 1a,b, two typical TC blister shapes, light and heavy TC blisters, can be observed in compressed SiAlNx films on glass substrates. Our experimental observation based on the optical images in Fig. 1a,b indicates that the projected area of the delamination zone of the TC blister is similar to the area swept by a segment of width 2b perpendicularly to a sinusoidal centreline. We then assume that the shape of the TC blister can thus be modelled as the postbuckling morphology of the thin plate clamped along such delamination boundary under equal biaxial residual compression (Fig. 1c). An approximate analytical solution for the shape of the TC blister is obtained in the following part. We argue that the shape of the TC blister can be characterized by several sectional profiles perpendicular to the centreline of the delamination zone (Fig. 1c). In contrast to the fact that the sectional profile of the straight-sided blister is always symmetric, the corresponding profile of the TC blister becomes asymmetric, dependent on the wavy amplitude of the centreline. The sequential sectional profiles of the TC blister exhibit a butterfly shape. Such geometric feature becomes more significant in the case of larger waviness amplitude of the centreline, corresponding to a transition from the light TC blister to the heavy TC blister. 2
a
b
b b
Centreline
c C-line
D-line m
z
x y
w
h
Centreline
s
B-line
b1
b2
Substrate
Figure 1 | Schematic model for the shape of telephone cord blisters. Fitting the projected area of (a) a light TC blister and (b) a heavy TC blister observed in SiAlNx films on glass substrates. (c) Sketch of the FvK model in a curvilinear coordinate under the clamping boundary condition along the boundary of the projected area. Scale bars in (a) and (b) are 30 and 50 mm, respectively.
Results Analytical solution to the 3D profile of TC blisters. We firstly determine the 3D shape of the TC blister by solving the FvK equations for a plate of thickness h clamped along the delamination boundary in a curvilinear coordinate (Fig. 1c). The centreline of the delamination zone is described by r0 ¼ ½x; A sinð2px=lÞ; 0, where A and l are the amplitude and the wavelength of the wavy centreline, respectively. The shape of the TC buckles reflected by the profile of the mid-plane of the thin plate after deformation can be expressed as r ¼ ½x ðx þ uÞ sin y; A sinð2px=lÞ þ ðx þ uÞ cos y; w
ð1Þ
where y (x) is defined as the angle between the tangential of the centreline and x axis, (s, x) is a curvilinear coordinate, u(s, x) is the displacement along x direction and w(s, x) is the deflection of the plate. Here we have assumed that the displacement change along the S direction is negligible compared to u(s, x) and w(s, x). The edges of the TC buckles are clamped at x ¼ ±b. We also assume that the in-plane displacement is small, that @u is, u x, u dy ds @x, and the out-of-plane deflection is relatively large. Based on the observed morphology of the TC buckles, we make another important assumption that u and w change @u @w @w slowly in the S direction, that is, @u @s @x and @s @x . We derive the FvK equations with the boundary condition in the curvilinear coordinate by minimizing the total elastic energy in the plate (see the details in the Supplementary Note 1 for the analytical solution to the TC buckle). The strain energy of the TC buckle within a period can be written as Z s0 Z b pffiffiffi ð2Þ F¼ f g d xds 0
b
where g is the determinant of the metric tensor gab and therefore pffiffiffi g dxds is the element of the area, f is the elastic strain energy
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
density per unit area having the form
1,200
3
ta0 ¼ 0
ð4Þ
f hpffiffigffi ð1 þ nÞem þ u0 þ 1 ðw0 Þ2 is the membrane where ta ¼ E 2 f ¼ E 2 is plane strain stress along the direction of x in the film. E 1n modulus with E, v the Young’s modulus and the Poisson’s ratio of the plate, respectively. em is the uniform equibiaxial residual strain in the plate. Equation (4) indicates the membrane stress ta should be a constant. Similarly, from dF=dw ¼ 0 we obtain the out-of-plane equilibrium equation of the plate " #0 2 00 00 y ;s w0 ¼ 0 ð5Þ 1 xy;s w lw0 þ 1 xy;s
D
1,000
C
B 800 z (nm)
h h f ¼ Dabrm gab grm þ Dabrm Dkab Dkrm ð3Þ 2 24 where gab, Dkab and Dabrm are defined as the mid-plane Lagrange strain tensor, the curvature change tensor due to deformation and the tensor of the elastic constants in the curvilinear coordinate, respectively. They can be explicitly expressed as a function of the variables s, x, u and w. Here the prime denotes differentiation with respect to x. If we only keep the leading order pffiffiffi term of u0 þ 12 ðw0 Þ2 , the element of area g 1 xy;s with 2 sinðqxÞ . The arc length of the centreline s(x) is y;s ¼ ½1 þA2Aq q2 cos2 ðqxÞ3=2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi given by ds ¼ 1 þ A2 q2 cos2 ðqxÞdx, with q ¼ 2p/l. We also assume that s ¼ 0 at x ¼ 0 and s ¼ s0 at x ¼ 2p/q. From dF=du ¼ 0, we have the in-plane equilibrium equation of the plate
600
B-line C-line
400
D-line Open symbol: Experiment Solid symbol: Simulation
200
Line: Theory for rigid substrate
0 4
8
12
16 (µm)
20
24
Figure 2 | Characterization of the TC blister shape. Several sectional profiles perpendicular to the centreline are shown in the inserted modified AFM image 40 mm 40 mm, which is updated by manually performing planefit offline of the original AFM image obtained by line-by-line (256 lines) scanning modes. The results from theory, simulation and AFM measurements are compared in a Ta film on glass substrates.
largest value in the profile along the B-line or D-line. This geometric feature of the TC blister is different from the previous report12 wherein the ridge line is viewed as a discontinuous contour line. Figure 2 demonstrates that the theoretical result 12ta where l ¼ Ef h3 . By solving equations (4) and (5) as an eigenvalue obtained by equation (6) predicts the asymmetric 3D feature problem, we get an approximate solution to the deflection of the of the TC blister, and the deflection amplitude variation of the TC ridge line. This is consistent with the AFM measurements plate up to the first order after the measured parameters of the buckled Ta film is given h
i
1 1 p p p as h ¼ 225 nm, b ¼ 10 mm, l/b ¼ 2, A/l ¼ 0.08, and the residual 1 y;s x 1 þ cos x þ y;s b2 x2 sin x wðx; xÞ w0 2 4 b 4b b strain in the Ta film is assumed to be em ¼ 0.004. The deflection ð6Þ amplitude of the ridge line given by equation (6) is underestimated because of the assumption of rigid substrate and where w0 is an Equation (6) is reduced to unknown p constant. clamping boundary condition. In fact, the substrate compliance 1 wðx; xÞ ¼ 2 w0 1 þ cos b x for the buckling of a straight strip could increase the maximum deflection26. We further numeriwith uniform width (y,s ¼ 0) (ref. 6). Therefore, w0 represents cally track the postbuckling morphology of the compressed plate the maximum deflection of the plate when y,s ¼ 0. Note that f h3 p2 on the substrate by setting the same predelamination zone, the E ta ¼ 12 b at y,s ¼ 0. Integrate this equation from b to b, same residual stress and the same other material parameters as and notice that u ¼ 0 at x ¼ ±b, one can get those in the case of the analytical solution. The film-to-substrate 640½12b2 em ð1 þ nÞ þ h2 p2 modulus ratio is set to be 2.5 (about 70 GPa for the glass substrate 2 w0 ¼ ð7Þ and 175 GPa for Ta film) based on the continuum model15. The 480p2 þ ð 465 110p2 þ 16p4 Þb2 y2;s numerical result in Fig. 2 matches the experimentally observed morphological feature of the TC blister even more closely. The characteristic shape of TC blisters. Equation (6) together The approximate solution in equation (6) also predicts the with equation (7) provides for the first time an approximate asymmetry denoted by b1/b2 and the variation of the maximum max analytical solution to the 3D profile of the TC buckle after the deflection at the ridge of the TC blister denoted by wmax Cline =wBline specific delamination area is available. To test whether the as a function of the waviness of the centreline denoted by A/l. In max solution can predict the 3D feature of the TC blister shape or the limit of A/l ¼ 0, b1/b2 ¼ 1 and wmax Cline =wBline ¼ 1, it is the not, Fig. 2 plots several sectional profiles of a buckled tantalum case of a straight-sided blister. The stability analysis for the (Ta) film on glass substrates perpendicular to the centreline straight-sided blister only predicted the onset stage during shown in the inserted modified AFM image with the size formation of the TC blister, namely, the amplitude of the of 40 mm 40 mm. The result in Fig. 2 shows that the sequence undulation is infinitely small corresponding to A/l-0 (ref. 8). of the sectional profiles of the TC blister perpendicular to the The current solution can apply to the case with finite value of A/l. centreline exhibits a butterfly shape reminiscent of the butterfly Figure 3 shows that different undulatory shapes of the TC blisters curve of strain versus applied load in shape memory alloys. can be described by two measurable geometrical parameters b1/b2 max The asymmetry is characterized by two different half-separations and wmax Cline =wBline as a function of A/l with the comparison denoted by b1 and b2 based on the ridge line of the TC blister between theory, simulation and experimental observations in (see Fig. 1c). The asymmetry has a maximum in the profile along various films including Ta, SiAlNx, Fe and Ni on glass substrates. max B-line or D-line, while the profile along the C-line is symmetric. With the increase of A/l, the values of b1/b2 and wmax Cline =wBline In addition, the maximum deflection in each sectional profile more significantly deviate from one, corresponding to the larger corresponding to the point at the ridge line is not constant. It asymmetry. However, we must note that the approximate has the smallest value in the profile along the C-line and the analytical solution is reliable only when the value of A/l is very NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
3
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
small since it proceeds by an asymptotic expansion in y,s (see the details in the Supplementary Note 1 for analytical solution to the TC buckle). From equation (5) it indicates that the solution may be questionable since there is a singularity as 1 xy;s ¼ 0. The singularity disappears when A=lo1=ð2p2 Þ ¼ 0:05 at l ¼ 2b. For the larger value of A/l, the numerical solution15 is more close to the experimental data, as shown in Fig. 3. It is found that the max variations of b1/b2 and wmax Cline =wBline tend to be saturated, respectively, after A/l40.08, in contrast to the monotonous deviation from one predicted by the approximate analytical solution. The result in Fig. 3 thus demonstrates that b1/b2 and max wmax Cline =wBline are another two measurable geometric parameters to characterize the undulatory morphology of the TC blister beside of l/b and A/l. The dependence of b1/b2 and max wmax Cline =wBline on A/l is attributed to the fact that the wavy centreline of the delamination zone breaks the symmetry
1.2
Theory Experiment Simulation
1.2
0.8
0.8
0.6
0.6
b1/b2
1.0
0.0
0.2
0.1
max w C-line /w max B-line
1.0
0.3
A/ max Figure 3 | Measurable geometrical parameters. b1/b2 and wmax Cline =wBline as a function of A/l characterize different undulatory shapes of the TC blisters: comparison between theory, simulation and experimental observations in various films including Ta, SiAlNx, Fe and Ni on glass substrates.
with respect to the straight centreline and introduces a position-dependent curvature as indicated in Supplementary Equation (10) in the buckled film wherein the buckling across the centreline may be not symmetric anymore. Light and heavy TC blisters. Figure 4 demonstrates that there are two kinds of buckle morphologies, light and heavy TC blisters, differentiated by the value of A/l. The upper row in Fig. 4 is the experimental observation in Ta and SiAlNx films on glass substrates, respectively. The lower row in Fig. 4 is the simulated postbuckling morphology given b/h ¼ 49, l/b ¼ 2 and em ¼ 0:005 at different values of A/l. It is found that with the increase of A/l, the buckle really turns out to be ubiquitously undulated from light to heavy TC blisters. The obtained result is consistent with the experimental observation shown in Supplementary Movie 1, where an initially straight blister progressively grows into the familiar TC structure. We believe that the transition point between the light and heavy TC blisters depends on the value of A/l. In the light TC blister, the value of A/l is very small, the width of the delamination area is close to 2b, the delamination boundary is smooth and the values of b1/b2 max and wmax Cline =wBline monotonously deviate from one. In contrast, in the heavy TC blister, the value of A/l is much larger, the delamination zone significantly widens, the values of b1/b2 max and wmax Cline =wBline are saturated, and the delamination boundary shows a cusp indicating the existence of a singularity (see Supplementary Fig. 3). In our analytical approach, l/b and A/l extracted from the projected area of the TC blister are assumed to be the input parameters instead of the prediction. Based on the assumption, max the equilibrium values of b1/b2 and wmax Cline =wBline that quantify the typical 3D features of the TC blister are determined by the postbuckling solution given delamination area. We can obtain the values of l/b and A/l from the experimental observation. The measurement of l/b is straightforward12. While the measurement of A/l could be obtained by tracking the undulation of the ridge line denoted by A0 =l, which value could be roughly viewed as that of A/l (Fig. 1a,b). The results
Light TC blister
Heavy TC blister
a
b
c
d
e
f
5 4 3 2 1 0
A/= 0.04
A/= 0.08
A/= 0.4
Figure 4 | Morphologies of light and heavy TC blisters. (a–c) AFM images. Scale bars are 10 mm. (d–f) Numerical simulation results with parameters b/h ¼ 49, l/b ¼ 2, em ¼ 0.005, and A/l ¼ 0.04, 0.08 and 0.4. 4
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
shown in Supplementary Fig. 2 indicate that both values of l/b and A0 =l in various observed TC blisters tend to be saturated and fall into a narrow range. The value of the former is between 1.6 and 2.8, consistent with the report in the literature16. The value of the latter is between 0.1 and 0.3. Since the TC blister can be developed from a secondary buckling instability of a straight-sided blister, where the instability mode predicts l/bE2 given the parameters b and A ¼ 0 (ref. 8), we believe that the waviness period of the TC blister may be inherited from the instability wavelength. The deviation from l/bE2 is attributed that l/b also depends on the adhesion with a non-trivial relationship14,16. Usually the increase of A/l releases more the elastic energy in the biaxially compressed film deposited on the substrate, while it has to increase the mixed-mode-dependent adhesion energy. The competition between them sets the equilibrium value of A/l. In addition, large value of A/l in the TC blisters is hard to report. It is possible because the case with a large value of A/l definitely increases the stress concentration which may lead to ridge crack20 or further buckling bifurcation, and the buckle is not TC blister any more15. Up to now, our modelling to the shape of the TC blister is obtained under the particular assumption on the shape of the delamination region. In fact, the delamination area is unnecessary to be sinusoidal shaped, especially during the oscillatory growth process of the TC blister11. Usually, determination of the buckle–delamination morphology can be separated into two steps. The first step is to derive a postbuckling solution given the delamination zone. The second step is to determine the delamination zone at equilibrium. If the edge of the delamination zone is described as the interfacial crack front, the delamination zone at equilibrium is determined as the buckling-mediated energy release rate is equal to the interface toughness. For the straight-sided blister, both steps are done analytically6. However, the analytical solution to the second step for the TC blister is not available. In fact, both the equilibrium delamination area and postbuckling morphology in thin films deposited on substrates can also be theoretically determined in principle. If we further take into account the contribution of the adhesion energy and the elastic strain energy in the substrate, all the equilibrium values max of l/b, A/l, b1/b2 and wmax Cline =wBline that quantify the TC blister shape with less restriction can be numerically determined by minimizing the total free energy in the film-substrate system. Our current numerical simulations rely on the recently developed continuum modelling to track the general morphological evolution of the buckle delamination without any restriction of the delamination zone15. In this approach, the concurrent buckling and delaminating processes are formulated using the time-dependent Ginzburg–Landau kinetic equations, driven by minimizing the film-substrate total free energy, including the elastic energies in both the film and the substrate, and the mixed-mode interfacial adhesion between them11,15. The effect of substrate elasticity and interfacial adhesion on the shape of the TC blister can be taken into account, which is neglected in the analytical approach. The modelling and simulation approach is outlined in Supplementary Note 3 and some results are shown in Supplementary Figs 4–6. The coupling behaviour between buckling and delamination in a film deposited on substrates with higher compressive stresses becomes more complex20,22, and the TC blister may exhibit beyond the sinusoidal configuration. Our numerical simulations not only recover the growth process of the TC blister from an initially circular blister7 but also capture a rich coalescence behaviour accompanied with the increase of the buckling width during further propagation of the TC buckle. The TC buckle becomes larger and larger with the appea-
rance of several spikes and/or daughter TC buckle at the outer undulated edge, consistent with our experimental observation (see Supplementary Fig. 6 and Supplementary Movies 2 and 3). Discussion In summary, the refined 3D morphological features of the TC buckle are elucidated by using AFM characterization, approximate analytical model and numerical simulations. We confirm that the shape of the TC blister can be modelled as the postbuckling morphology of the compressed plate clamped along the delamination front given the parameters l/b and A/l. Two measurable geometrical parameters b1/b2 and max wmax Cline =wBline are proposed to characterize a so-called ‘butterfly shape’ of the sequential sectional profiles, which are universal in both the light and heavy TC blisters. The above features are successfully reproduced by our approximate analytical model and numerical simulations. Furthermore, how the fully nonlinear buckle-driven delamination process leads to the morphological evolution of the TC blister is captured by numerical simulations and experimental observation. The present work provides insight into the 3D shape of TC buckles. Methods Formation of the TC buckle as a result of the buckle-driven delamination process was simulated using the phase field method where the film buckles into an equilibrium buckle–delamination configuration driven by minimizing the total free energy. The total free energy of the film-substrate system is established by incorporating Green function method for the substrate elasticity, FvK plate theory for nonlinear film deformation and cohesive zone model for mixed-mode interfacial adhesion. The total free energy of the film-substrate system including the film, the substrate and the interface can be expressed as a functional of the out-ofplane displacement of the film z(x, t) and the displacement jump vector across the interface Lðx; t Þ as U tot ¼ U film þ U sub þ U int ; where Ufilm is the elastic strain energy including the bending and stretching energies in the film, Usub is the elastic energy of the substrate and Uint is the adhesion energy between the film and the substrate. Their detailed expressions are found in ref. 15. Following ref. 11, we may also adopt alternative cohesive zone model with bilinear traction versus separation law for better description of mixed-mode dependence of interface adhesion. The dynamic equations for z; Li denoted as the time-dependent Ginzburg–Landau kinetic equations to describe the minimization process of the total free energy are given by: @z=@t ¼ Gz dU tot =dz and @Li =@t ¼ GLi dU tot =dLi , where t denotes time and Gz , GLi are the kinetic coefficients that characterize the relaxation rates of the buckling and delamination processes in the overdamped dynamics. This gradient flow form of the dynamics guarantees a monotonous minimization process of Utot and ensures a convergent solution of the dynamic equations, whose steady-state solutions provide the configuration at the equilibrium. We solve the reduced forms in a computational cell with periodic boundary conditions using input parameters: e0ab , mf =ms , vf ¼ 0:3, vs ¼ 0:5, GLi ¼ GLi =Gz ¼ 0:1, dn ¼ dt ¼ 0:2h, gn ¼ gn =edn ms and gt ¼ gt =edn ms . The grid spacing is Dx ¼ Dy ¼ h and the time step is Dt ¼ 0:1t=t, with t ¼ h=Gz ms . A small random fluctuation mimicking the thermal fluctuation is used to facilitate nucleation of buckling delamination from the pre-existing interfacial delamination region, in which zero interface toughness is assumed.
Data availability. The data that support the findings of this study are available from the corresponding author on request.
References 1. Hutchinson, J. W. & Koiter, W. T. Postbuckling theory. App. Mech. Rev. 23, 1353–1366 (1970). 2. Pocivavsek, L. et al. Stress and fold localization in thin elastic membranes. Science 320, 912–916 (2008). 3. Brau, F. et al. Multiple-length-scale elastic instability mimics parametric resonance of nonlinear oscillators. Nat. Phys. 7, 56–60 (2011). 4. Freund, L. B. & Suresh, S Thin Film Materials: Stress, Defect Formation and Surface Evolution (Cambridge University Press, 2003). 5. Gioia, G. & Ortiz, M. Delamination of compressed thin films. Adv. Appl. Mech. 33, 119–192 (1997). 6. Hutchinson, J. W. & Suo, Z. Mixed-mode cracking in layered materials. Adv. Appl. Mech. 29, 63–191 (1992).
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
5
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
7. Hutchinson, J. W., Thouless, M. D. & Liniger, E. G. Growth and configurational stability of circular, buckling-driven dilm delaminations. Acta Metall. Mater. 40, 295–308 (1992). 8. Audoly, B. Stability of straight delamination blisters. Phys. Rev. Lett. 83, 4124–4127 (1999). 9. Parry, G., Cimetiere, A., Coupeau, C., Colin, J. & Grilhe, J. Stability diagram of unilateral buckling patterns of strip-delaminated films. Phys. Rev. E 74, 066601 (2006). 10. Thouless, M. D. Combined buckling and cracking of films. J. Am. Ceram. Soc. 76, 2936–2938 (1993). 11. Faou, J. Y., Parry, G., Grachev, S. & Barthel, E. How does adhesion induce the formation of telephone cord buckles? Phys. Rev. Lett. 108, 116102 (2012). 12. Moon, M. W., Jensen, H. M., Hutchinson, J. W., Oh, K. H. & Evans, A. G. The characterization of telephone cord buckling of compressed thin films on substrates. J. Mech. Phys. Solids 50, 2355–2377 (2002). 13. Moon, M. W., Lee, K. R., Oh, K. H. & Hutchinson, J. W. Buckle delamination on patterned substrates. Acta Mater. 52, 3151–3159 (2004). 14. Pan, K., Ni, Y. & He, L. H. Effects of interface sliding on the formation of telephone cord buckles. Phys. Rev. E 88, 062405 (2013). 15. Ni, Y. & Soh, A. K. On the growth of buckle-delamination pattern in compressed anisotropic thin films. Acta Mater. 69, 37–46 (2014). 16. Faou, J. Y., Parry, G., Grachev, S. & Barthel, E. Telephone cord buckles—a relation between wavelength and adhesion. J. Mech. Phys. Solids 75, 93–103 (2015). 17. Crosby, K. M. & Bradley, R. M. Pattern formation during delamination and buckling of thin films. Phys. Rev. E 59, R2542 (1999). 18. Peyla, P. Undulated blistering during thin film delamination. Phys. Rev. E 62, R1501 (2000). 19. Jagla, E. A. Modeling the buckling and delamination of thin films. Phys. Rev. B 75, 085405 (2007). 20. Yu, S. J. et al. Morphological selections and dynamical evolutions of buckling patterns in SiAlNx films: from straight-sided to telephone cord or bubble structures. Acta Mater. 64, 41–53 (2014). 21. Liu, Y. P. et al. Large-area, periodic, hexagonal wrinkles on nanocrystalline graphitic film. Adv. Funct. Mater. 25, 5492–5503 (2015). 22. Yu, S. J. et al. Spatial and kinetic evolutions of telephone cord buckles. Surf. Coat. Technol. 228, 258–265 (2013). 23. Jin, W. M. & Sternberga, P. Energy estimates for the von Ka´rma´n model of thin-film blistering. J. Math. Phys. 42, 192–199 (2001). 24. Angelillo, M., Babilio, E., Cardamone, L., Fortunato, A. & Lippiello, M. Some remarks on the retrofitting of masonry structures with composite materials. Compos Part B 61, 11–16 (2014). 25. Belgacem, H. B., Conti, S., DeSimone, A. & Mu¨ller, S. Energy scaling of compressed elastic films—three-dimensional elasticity and reduced theories. Arch. Ration. Mech. Anal. 164, 1–37 (2002).
6
26. Parry, G. et al. Effect of substrate compliance on the global unilateral postbuckling of coatings: AFM observations and finite element calculations. Acta Mater. 53, 441–447 (2005).
Acknowledgements This work was supported by the National Natural Science Foundation of China (Grant Nos 11222219, 11472262, 11132009, 11472271, 11622222 and 11204283), the Collaborative Innovation Center of Suzhou Nano Science and Technology, the Fundamental Research Funds for the Central Universities (Grant Nos WK2090050027 and WK2480000001) and the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB22040502 and XDB22040403).
Author contributions Y. Ni, S. Yu, H. Jiang and L. He initiated and supervised the project. H. Jiang conceived the theory. Y. Ni performed phase field simulation. S. Yu performed AFM measurement. All authors analysed the data and wrote the manuscript.
Additional information Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications Competing financial interests: The authors declare no competing financial interests. Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/ How to cite this article: Ni, Y. et al. The shape of telephone cord blisters. Nat. Commun. 8, 14138 doi: 10.1038/ncomms14138 (2017). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
r The Author(s) 2017
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
DOI: 10.1038/ncomms14138
OPEN
The shape of telephone cord blisters Yong Ni1, Senjiang Yu2, Hongyuan Jiang1 & Linghui He1
Formation of telephone cord blisters as a result of buckling delamination is widely observed in many compressed film-substrate systems. Here we report a universal morphological feature of such blisters characterized by their sequential sectional profiles exhibiting a butterfly shape using atomic force microscopy. Two kinds of buckle morphologies, light and heavy telephone cord blisters, are observed and differentiated by measurable geometrical parameters. Based on the Fo¨ppl-von Ka´rma´n plate theory, the observed three-dimensional features of the telephone cord blister are predicted by the proposed approximate analytical model and simulation. The latter further replicates growth and coalescence of the telephone cord into complex buckling delamination patterns observed in the experiment.
1 CAS
Key Laboratory of Mechanical Behavior and Design of Materials, Department of Modern Mechanics, University of Science and Technology of China, Hefei, Anhui 230026, China. 2 Department of Physics, China Jiliang University, Hangzhou 310018, China. Correspondence and requests for materials should be addressed to H.J. (email: [email protected]). NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
1
ARTICLE
D
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
etailed origins of postbuckling phenomena have been a subject of discussions for several decades1. The buckles usually involve multiple instabilities and exhibit intriguing morphologies2,3. When a film deposited on a substrate is subject to a large residual compression, it buckles and delaminates away from the substrate, and the blisters with straight-sided, circular, telephone cord (TC) or networklike patterns are formed4,5. Predicting the shape of such blisters is of scientific interest for stress-driven pattern formation and important for avoiding structural failure in various film-substrate systems4. Among various manifestations of this phenomenon, the TC blister is the most frequently observed in many different film-substrate systems5. This motivates a search to identify universal features of TC blisters. Recently, TC formation has been considered as a development from a secondary instability of straight-sided or circular blisters. The instability condition has been analytically addressed by the Fo¨ppl-von Ka´rma´n (FvK) plate model6–9. Further, it has been found that the TC blister can also grow with the appearance of sags by an oscillatory process and its edge unnecessarily starts straight10,11. So far, a convincing solution for the three-dimensional (3D) TC blister morphology remains a challenge because of limited methods for handling nonlinear elasticity of thin plate, although a series of numerical models and experiments have provided valuable insights into the morphological features of TC blisters10–25. The elegant instability analysis of a straight-sided blister predicted a constant undulation-period-to-width ratio of the TC buckle under the clamping boundary conditions close to the experimental observation8. In addition, the ratio may be not constant and change with the interfacial adhesion and the film-to-substrate modulus ratio11–16. The zigzag undulation feature of the TC buckle has been replicated by numerical investigations5,15–19. A comprehensive study for the 3D TC morphology suggested that it could be approximately viewed as a sequence of connected segments of a circular buckle pinned at its centre12. However, such approximation predicted the ridge of the TC blister to be discontinuous, different from our following refined experimental observations wherein the ridge is continuous and its height is periodically changing. Here we aim to probe the 3D features of the TC morphology using a combination of analytical model, numerical simulation and atomic force microscopy (AFM) study. As shown in Fig. 1a,b, two typical TC blister shapes, light and heavy TC blisters, can be observed in compressed SiAlNx films on glass substrates. Our experimental observation based on the optical images in Fig. 1a,b indicates that the projected area of the delamination zone of the TC blister is similar to the area swept by a segment of width 2b perpendicularly to a sinusoidal centreline. We then assume that the shape of the TC blister can thus be modelled as the postbuckling morphology of the thin plate clamped along such delamination boundary under equal biaxial residual compression (Fig. 1c). An approximate analytical solution for the shape of the TC blister is obtained in the following part. We argue that the shape of the TC blister can be characterized by several sectional profiles perpendicular to the centreline of the delamination zone (Fig. 1c). In contrast to the fact that the sectional profile of the straight-sided blister is always symmetric, the corresponding profile of the TC blister becomes asymmetric, dependent on the wavy amplitude of the centreline. The sequential sectional profiles of the TC blister exhibit a butterfly shape. Such geometric feature becomes more significant in the case of larger waviness amplitude of the centreline, corresponding to a transition from the light TC blister to the heavy TC blister. 2
a
b
b b
Centreline
c C-line
D-line m
z
x y
w
h
Centreline
s
B-line
b1
b2
Substrate
Figure 1 | Schematic model for the shape of telephone cord blisters. Fitting the projected area of (a) a light TC blister and (b) a heavy TC blister observed in SiAlNx films on glass substrates. (c) Sketch of the FvK model in a curvilinear coordinate under the clamping boundary condition along the boundary of the projected area. Scale bars in (a) and (b) are 30 and 50 mm, respectively.
Results Analytical solution to the 3D profile of TC blisters. We firstly determine the 3D shape of the TC blister by solving the FvK equations for a plate of thickness h clamped along the delamination boundary in a curvilinear coordinate (Fig. 1c). The centreline of the delamination zone is described by r0 ¼ ½x; A sinð2px=lÞ; 0, where A and l are the amplitude and the wavelength of the wavy centreline, respectively. The shape of the TC buckles reflected by the profile of the mid-plane of the thin plate after deformation can be expressed as r ¼ ½x ðx þ uÞ sin y; A sinð2px=lÞ þ ðx þ uÞ cos y; w
ð1Þ
where y (x) is defined as the angle between the tangential of the centreline and x axis, (s, x) is a curvilinear coordinate, u(s, x) is the displacement along x direction and w(s, x) is the deflection of the plate. Here we have assumed that the displacement change along the S direction is negligible compared to u(s, x) and w(s, x). The edges of the TC buckles are clamped at x ¼ ±b. We also assume that the in-plane displacement is small, that @u is, u x, u dy ds @x, and the out-of-plane deflection is relatively large. Based on the observed morphology of the TC buckles, we make another important assumption that u and w change @u @w @w slowly in the S direction, that is, @u @s @x and @s @x . We derive the FvK equations with the boundary condition in the curvilinear coordinate by minimizing the total elastic energy in the plate (see the details in the Supplementary Note 1 for the analytical solution to the TC buckle). The strain energy of the TC buckle within a period can be written as Z s0 Z b pffiffiffi ð2Þ F¼ f g d xds 0
b
where g is the determinant of the metric tensor gab and therefore pffiffiffi g dxds is the element of the area, f is the elastic strain energy
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
density per unit area having the form
1,200
3
ta0 ¼ 0
ð4Þ
f hpffiffigffi ð1 þ nÞem þ u0 þ 1 ðw0 Þ2 is the membrane where ta ¼ E 2 f ¼ E 2 is plane strain stress along the direction of x in the film. E 1n modulus with E, v the Young’s modulus and the Poisson’s ratio of the plate, respectively. em is the uniform equibiaxial residual strain in the plate. Equation (4) indicates the membrane stress ta should be a constant. Similarly, from dF=dw ¼ 0 we obtain the out-of-plane equilibrium equation of the plate " #0 2 00 00 y ;s w0 ¼ 0 ð5Þ 1 xy;s w lw0 þ 1 xy;s
D
1,000
C
B 800 z (nm)
h h f ¼ Dabrm gab grm þ Dabrm Dkab Dkrm ð3Þ 2 24 where gab, Dkab and Dabrm are defined as the mid-plane Lagrange strain tensor, the curvature change tensor due to deformation and the tensor of the elastic constants in the curvilinear coordinate, respectively. They can be explicitly expressed as a function of the variables s, x, u and w. Here the prime denotes differentiation with respect to x. If we only keep the leading order pffiffiffi term of u0 þ 12 ðw0 Þ2 , the element of area g 1 xy;s with 2 sinðqxÞ . The arc length of the centreline s(x) is y;s ¼ ½1 þA2Aq q2 cos2 ðqxÞ3=2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi given by ds ¼ 1 þ A2 q2 cos2 ðqxÞdx, with q ¼ 2p/l. We also assume that s ¼ 0 at x ¼ 0 and s ¼ s0 at x ¼ 2p/q. From dF=du ¼ 0, we have the in-plane equilibrium equation of the plate
600
B-line C-line
400
D-line Open symbol: Experiment Solid symbol: Simulation
200
Line: Theory for rigid substrate
0 4
8
12
16 (µm)
20
24
Figure 2 | Characterization of the TC blister shape. Several sectional profiles perpendicular to the centreline are shown in the inserted modified AFM image 40 mm 40 mm, which is updated by manually performing planefit offline of the original AFM image obtained by line-by-line (256 lines) scanning modes. The results from theory, simulation and AFM measurements are compared in a Ta film on glass substrates.
largest value in the profile along the B-line or D-line. This geometric feature of the TC blister is different from the previous report12 wherein the ridge line is viewed as a discontinuous contour line. Figure 2 demonstrates that the theoretical result 12ta where l ¼ Ef h3 . By solving equations (4) and (5) as an eigenvalue obtained by equation (6) predicts the asymmetric 3D feature problem, we get an approximate solution to the deflection of the of the TC blister, and the deflection amplitude variation of the TC ridge line. This is consistent with the AFM measurements plate up to the first order after the measured parameters of the buckled Ta film is given h
i
1 1 p p p as h ¼ 225 nm, b ¼ 10 mm, l/b ¼ 2, A/l ¼ 0.08, and the residual 1 y;s x 1 þ cos x þ y;s b2 x2 sin x wðx; xÞ w0 2 4 b 4b b strain in the Ta film is assumed to be em ¼ 0.004. The deflection ð6Þ amplitude of the ridge line given by equation (6) is underestimated because of the assumption of rigid substrate and where w0 is an Equation (6) is reduced to unknown p constant. clamping boundary condition. In fact, the substrate compliance 1 wðx; xÞ ¼ 2 w0 1 þ cos b x for the buckling of a straight strip could increase the maximum deflection26. We further numeriwith uniform width (y,s ¼ 0) (ref. 6). Therefore, w0 represents cally track the postbuckling morphology of the compressed plate the maximum deflection of the plate when y,s ¼ 0. Note that f h3 p2 on the substrate by setting the same predelamination zone, the E ta ¼ 12 b at y,s ¼ 0. Integrate this equation from b to b, same residual stress and the same other material parameters as and notice that u ¼ 0 at x ¼ ±b, one can get those in the case of the analytical solution. The film-to-substrate 640½12b2 em ð1 þ nÞ þ h2 p2 modulus ratio is set to be 2.5 (about 70 GPa for the glass substrate 2 w0 ¼ ð7Þ and 175 GPa for Ta film) based on the continuum model15. The 480p2 þ ð 465 110p2 þ 16p4 Þb2 y2;s numerical result in Fig. 2 matches the experimentally observed morphological feature of the TC blister even more closely. The characteristic shape of TC blisters. Equation (6) together The approximate solution in equation (6) also predicts the with equation (7) provides for the first time an approximate asymmetry denoted by b1/b2 and the variation of the maximum max analytical solution to the 3D profile of the TC buckle after the deflection at the ridge of the TC blister denoted by wmax Cline =wBline specific delamination area is available. To test whether the as a function of the waviness of the centreline denoted by A/l. In max solution can predict the 3D feature of the TC blister shape or the limit of A/l ¼ 0, b1/b2 ¼ 1 and wmax Cline =wBline ¼ 1, it is the not, Fig. 2 plots several sectional profiles of a buckled tantalum case of a straight-sided blister. The stability analysis for the (Ta) film on glass substrates perpendicular to the centreline straight-sided blister only predicted the onset stage during shown in the inserted modified AFM image with the size formation of the TC blister, namely, the amplitude of the of 40 mm 40 mm. The result in Fig. 2 shows that the sequence undulation is infinitely small corresponding to A/l-0 (ref. 8). of the sectional profiles of the TC blister perpendicular to the The current solution can apply to the case with finite value of A/l. centreline exhibits a butterfly shape reminiscent of the butterfly Figure 3 shows that different undulatory shapes of the TC blisters curve of strain versus applied load in shape memory alloys. can be described by two measurable geometrical parameters b1/b2 max The asymmetry is characterized by two different half-separations and wmax Cline =wBline as a function of A/l with the comparison denoted by b1 and b2 based on the ridge line of the TC blister between theory, simulation and experimental observations in (see Fig. 1c). The asymmetry has a maximum in the profile along various films including Ta, SiAlNx, Fe and Ni on glass substrates. max B-line or D-line, while the profile along the C-line is symmetric. With the increase of A/l, the values of b1/b2 and wmax Cline =wBline In addition, the maximum deflection in each sectional profile more significantly deviate from one, corresponding to the larger corresponding to the point at the ridge line is not constant. It asymmetry. However, we must note that the approximate has the smallest value in the profile along the C-line and the analytical solution is reliable only when the value of A/l is very NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
3
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
small since it proceeds by an asymptotic expansion in y,s (see the details in the Supplementary Note 1 for analytical solution to the TC buckle). From equation (5) it indicates that the solution may be questionable since there is a singularity as 1 xy;s ¼ 0. The singularity disappears when A=lo1=ð2p2 Þ ¼ 0:05 at l ¼ 2b. For the larger value of A/l, the numerical solution15 is more close to the experimental data, as shown in Fig. 3. It is found that the max variations of b1/b2 and wmax Cline =wBline tend to be saturated, respectively, after A/l40.08, in contrast to the monotonous deviation from one predicted by the approximate analytical solution. The result in Fig. 3 thus demonstrates that b1/b2 and max wmax Cline =wBline are another two measurable geometric parameters to characterize the undulatory morphology of the TC blister beside of l/b and A/l. The dependence of b1/b2 and max wmax Cline =wBline on A/l is attributed to the fact that the wavy centreline of the delamination zone breaks the symmetry
1.2
Theory Experiment Simulation
1.2
0.8
0.8
0.6
0.6
b1/b2
1.0
0.0
0.2
0.1
max w C-line /w max B-line
1.0
0.3
A/ max Figure 3 | Measurable geometrical parameters. b1/b2 and wmax Cline =wBline as a function of A/l characterize different undulatory shapes of the TC blisters: comparison between theory, simulation and experimental observations in various films including Ta, SiAlNx, Fe and Ni on glass substrates.
with respect to the straight centreline and introduces a position-dependent curvature as indicated in Supplementary Equation (10) in the buckled film wherein the buckling across the centreline may be not symmetric anymore. Light and heavy TC blisters. Figure 4 demonstrates that there are two kinds of buckle morphologies, light and heavy TC blisters, differentiated by the value of A/l. The upper row in Fig. 4 is the experimental observation in Ta and SiAlNx films on glass substrates, respectively. The lower row in Fig. 4 is the simulated postbuckling morphology given b/h ¼ 49, l/b ¼ 2 and em ¼ 0:005 at different values of A/l. It is found that with the increase of A/l, the buckle really turns out to be ubiquitously undulated from light to heavy TC blisters. The obtained result is consistent with the experimental observation shown in Supplementary Movie 1, where an initially straight blister progressively grows into the familiar TC structure. We believe that the transition point between the light and heavy TC blisters depends on the value of A/l. In the light TC blister, the value of A/l is very small, the width of the delamination area is close to 2b, the delamination boundary is smooth and the values of b1/b2 max and wmax Cline =wBline monotonously deviate from one. In contrast, in the heavy TC blister, the value of A/l is much larger, the delamination zone significantly widens, the values of b1/b2 max and wmax Cline =wBline are saturated, and the delamination boundary shows a cusp indicating the existence of a singularity (see Supplementary Fig. 3). In our analytical approach, l/b and A/l extracted from the projected area of the TC blister are assumed to be the input parameters instead of the prediction. Based on the assumption, max the equilibrium values of b1/b2 and wmax Cline =wBline that quantify the typical 3D features of the TC blister are determined by the postbuckling solution given delamination area. We can obtain the values of l/b and A/l from the experimental observation. The measurement of l/b is straightforward12. While the measurement of A/l could be obtained by tracking the undulation of the ridge line denoted by A0 =l, which value could be roughly viewed as that of A/l (Fig. 1a,b). The results
Light TC blister
Heavy TC blister
a
b
c
d
e
f
5 4 3 2 1 0
A/= 0.04
A/= 0.08
A/= 0.4
Figure 4 | Morphologies of light and heavy TC blisters. (a–c) AFM images. Scale bars are 10 mm. (d–f) Numerical simulation results with parameters b/h ¼ 49, l/b ¼ 2, em ¼ 0.005, and A/l ¼ 0.04, 0.08 and 0.4. 4
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
shown in Supplementary Fig. 2 indicate that both values of l/b and A0 =l in various observed TC blisters tend to be saturated and fall into a narrow range. The value of the former is between 1.6 and 2.8, consistent with the report in the literature16. The value of the latter is between 0.1 and 0.3. Since the TC blister can be developed from a secondary buckling instability of a straight-sided blister, where the instability mode predicts l/bE2 given the parameters b and A ¼ 0 (ref. 8), we believe that the waviness period of the TC blister may be inherited from the instability wavelength. The deviation from l/bE2 is attributed that l/b also depends on the adhesion with a non-trivial relationship14,16. Usually the increase of A/l releases more the elastic energy in the biaxially compressed film deposited on the substrate, while it has to increase the mixed-mode-dependent adhesion energy. The competition between them sets the equilibrium value of A/l. In addition, large value of A/l in the TC blisters is hard to report. It is possible because the case with a large value of A/l definitely increases the stress concentration which may lead to ridge crack20 or further buckling bifurcation, and the buckle is not TC blister any more15. Up to now, our modelling to the shape of the TC blister is obtained under the particular assumption on the shape of the delamination region. In fact, the delamination area is unnecessary to be sinusoidal shaped, especially during the oscillatory growth process of the TC blister11. Usually, determination of the buckle–delamination morphology can be separated into two steps. The first step is to derive a postbuckling solution given the delamination zone. The second step is to determine the delamination zone at equilibrium. If the edge of the delamination zone is described as the interfacial crack front, the delamination zone at equilibrium is determined as the buckling-mediated energy release rate is equal to the interface toughness. For the straight-sided blister, both steps are done analytically6. However, the analytical solution to the second step for the TC blister is not available. In fact, both the equilibrium delamination area and postbuckling morphology in thin films deposited on substrates can also be theoretically determined in principle. If we further take into account the contribution of the adhesion energy and the elastic strain energy in the substrate, all the equilibrium values max of l/b, A/l, b1/b2 and wmax Cline =wBline that quantify the TC blister shape with less restriction can be numerically determined by minimizing the total free energy in the film-substrate system. Our current numerical simulations rely on the recently developed continuum modelling to track the general morphological evolution of the buckle delamination without any restriction of the delamination zone15. In this approach, the concurrent buckling and delaminating processes are formulated using the time-dependent Ginzburg–Landau kinetic equations, driven by minimizing the film-substrate total free energy, including the elastic energies in both the film and the substrate, and the mixed-mode interfacial adhesion between them11,15. The effect of substrate elasticity and interfacial adhesion on the shape of the TC blister can be taken into account, which is neglected in the analytical approach. The modelling and simulation approach is outlined in Supplementary Note 3 and some results are shown in Supplementary Figs 4–6. The coupling behaviour between buckling and delamination in a film deposited on substrates with higher compressive stresses becomes more complex20,22, and the TC blister may exhibit beyond the sinusoidal configuration. Our numerical simulations not only recover the growth process of the TC blister from an initially circular blister7 but also capture a rich coalescence behaviour accompanied with the increase of the buckling width during further propagation of the TC buckle. The TC buckle becomes larger and larger with the appea-
rance of several spikes and/or daughter TC buckle at the outer undulated edge, consistent with our experimental observation (see Supplementary Fig. 6 and Supplementary Movies 2 and 3). Discussion In summary, the refined 3D morphological features of the TC buckle are elucidated by using AFM characterization, approximate analytical model and numerical simulations. We confirm that the shape of the TC blister can be modelled as the postbuckling morphology of the compressed plate clamped along the delamination front given the parameters l/b and A/l. Two measurable geometrical parameters b1/b2 and max wmax Cline =wBline are proposed to characterize a so-called ‘butterfly shape’ of the sequential sectional profiles, which are universal in both the light and heavy TC blisters. The above features are successfully reproduced by our approximate analytical model and numerical simulations. Furthermore, how the fully nonlinear buckle-driven delamination process leads to the morphological evolution of the TC blister is captured by numerical simulations and experimental observation. The present work provides insight into the 3D shape of TC buckles. Methods Formation of the TC buckle as a result of the buckle-driven delamination process was simulated using the phase field method where the film buckles into an equilibrium buckle–delamination configuration driven by minimizing the total free energy. The total free energy of the film-substrate system is established by incorporating Green function method for the substrate elasticity, FvK plate theory for nonlinear film deformation and cohesive zone model for mixed-mode interfacial adhesion. The total free energy of the film-substrate system including the film, the substrate and the interface can be expressed as a functional of the out-ofplane displacement of the film z(x, t) and the displacement jump vector across the interface Lðx; t Þ as U tot ¼ U film þ U sub þ U int ; where Ufilm is the elastic strain energy including the bending and stretching energies in the film, Usub is the elastic energy of the substrate and Uint is the adhesion energy between the film and the substrate. Their detailed expressions are found in ref. 15. Following ref. 11, we may also adopt alternative cohesive zone model with bilinear traction versus separation law for better description of mixed-mode dependence of interface adhesion. The dynamic equations for z; Li denoted as the time-dependent Ginzburg–Landau kinetic equations to describe the minimization process of the total free energy are given by: @z=@t ¼ Gz dU tot =dz and @Li =@t ¼ GLi dU tot =dLi , where t denotes time and Gz , GLi are the kinetic coefficients that characterize the relaxation rates of the buckling and delamination processes in the overdamped dynamics. This gradient flow form of the dynamics guarantees a monotonous minimization process of Utot and ensures a convergent solution of the dynamic equations, whose steady-state solutions provide the configuration at the equilibrium. We solve the reduced forms in a computational cell with periodic boundary conditions using input parameters: e0ab , mf =ms , vf ¼ 0:3, vs ¼ 0:5, GLi ¼ GLi =Gz ¼ 0:1, dn ¼ dt ¼ 0:2h, gn ¼ gn =edn ms and gt ¼ gt =edn ms . The grid spacing is Dx ¼ Dy ¼ h and the time step is Dt ¼ 0:1t=t, with t ¼ h=Gz ms . A small random fluctuation mimicking the thermal fluctuation is used to facilitate nucleation of buckling delamination from the pre-existing interfacial delamination region, in which zero interface toughness is assumed.
Data availability. The data that support the findings of this study are available from the corresponding author on request.
References 1. Hutchinson, J. W. & Koiter, W. T. Postbuckling theory. App. Mech. Rev. 23, 1353–1366 (1970). 2. Pocivavsek, L. et al. Stress and fold localization in thin elastic membranes. Science 320, 912–916 (2008). 3. Brau, F. et al. Multiple-length-scale elastic instability mimics parametric resonance of nonlinear oscillators. Nat. Phys. 7, 56–60 (2011). 4. Freund, L. B. & Suresh, S Thin Film Materials: Stress, Defect Formation and Surface Evolution (Cambridge University Press, 2003). 5. Gioia, G. & Ortiz, M. Delamination of compressed thin films. Adv. Appl. Mech. 33, 119–192 (1997). 6. Hutchinson, J. W. & Suo, Z. Mixed-mode cracking in layered materials. Adv. Appl. Mech. 29, 63–191 (1992).
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
5
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14138
7. Hutchinson, J. W., Thouless, M. D. & Liniger, E. G. Growth and configurational stability of circular, buckling-driven dilm delaminations. Acta Metall. Mater. 40, 295–308 (1992). 8. Audoly, B. Stability of straight delamination blisters. Phys. Rev. Lett. 83, 4124–4127 (1999). 9. Parry, G., Cimetiere, A., Coupeau, C., Colin, J. & Grilhe, J. Stability diagram of unilateral buckling patterns of strip-delaminated films. Phys. Rev. E 74, 066601 (2006). 10. Thouless, M. D. Combined buckling and cracking of films. J. Am. Ceram. Soc. 76, 2936–2938 (1993). 11. Faou, J. Y., Parry, G., Grachev, S. & Barthel, E. How does adhesion induce the formation of telephone cord buckles? Phys. Rev. Lett. 108, 116102 (2012). 12. Moon, M. W., Jensen, H. M., Hutchinson, J. W., Oh, K. H. & Evans, A. G. The characterization of telephone cord buckling of compressed thin films on substrates. J. Mech. Phys. Solids 50, 2355–2377 (2002). 13. Moon, M. W., Lee, K. R., Oh, K. H. & Hutchinson, J. W. Buckle delamination on patterned substrates. Acta Mater. 52, 3151–3159 (2004). 14. Pan, K., Ni, Y. & He, L. H. Effects of interface sliding on the formation of telephone cord buckles. Phys. Rev. E 88, 062405 (2013). 15. Ni, Y. & Soh, A. K. On the growth of buckle-delamination pattern in compressed anisotropic thin films. Acta Mater. 69, 37–46 (2014). 16. Faou, J. Y., Parry, G., Grachev, S. & Barthel, E. Telephone cord buckles—a relation between wavelength and adhesion. J. Mech. Phys. Solids 75, 93–103 (2015). 17. Crosby, K. M. & Bradley, R. M. Pattern formation during delamination and buckling of thin films. Phys. Rev. E 59, R2542 (1999). 18. Peyla, P. Undulated blistering during thin film delamination. Phys. Rev. E 62, R1501 (2000). 19. Jagla, E. A. Modeling the buckling and delamination of thin films. Phys. Rev. B 75, 085405 (2007). 20. Yu, S. J. et al. Morphological selections and dynamical evolutions of buckling patterns in SiAlNx films: from straight-sided to telephone cord or bubble structures. Acta Mater. 64, 41–53 (2014). 21. Liu, Y. P. et al. Large-area, periodic, hexagonal wrinkles on nanocrystalline graphitic film. Adv. Funct. Mater. 25, 5492–5503 (2015). 22. Yu, S. J. et al. Spatial and kinetic evolutions of telephone cord buckles. Surf. Coat. Technol. 228, 258–265 (2013). 23. Jin, W. M. & Sternberga, P. Energy estimates for the von Ka´rma´n model of thin-film blistering. J. Math. Phys. 42, 192–199 (2001). 24. Angelillo, M., Babilio, E., Cardamone, L., Fortunato, A. & Lippiello, M. Some remarks on the retrofitting of masonry structures with composite materials. Compos Part B 61, 11–16 (2014). 25. Belgacem, H. B., Conti, S., DeSimone, A. & Mu¨ller, S. Energy scaling of compressed elastic films—three-dimensional elasticity and reduced theories. Arch. Ration. Mech. Anal. 164, 1–37 (2002).
6
26. Parry, G. et al. Effect of substrate compliance on the global unilateral postbuckling of coatings: AFM observations and finite element calculations. Acta Mater. 53, 441–447 (2005).
Acknowledgements This work was supported by the National Natural Science Foundation of China (Grant Nos 11222219, 11472262, 11132009, 11472271, 11622222 and 11204283), the Collaborative Innovation Center of Suzhou Nano Science and Technology, the Fundamental Research Funds for the Central Universities (Grant Nos WK2090050027 and WK2480000001) and the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB22040502 and XDB22040403).
Author contributions Y. Ni, S. Yu, H. Jiang and L. He initiated and supervised the project. H. Jiang conceived the theory. Y. Ni performed phase field simulation. S. Yu performed AFM measurement. All authors analysed the data and wrote the manuscript.
Additional information Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications Competing financial interests: The authors declare no competing financial interests. Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/ How to cite this article: Ni, Y. et al. The shape of telephone cord blisters. Nat. Commun. 8, 14138 doi: 10.1038/ncomms14138 (2017). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
r The Author(s) 2017
NATURE COMMUNICATIONS | 8:14138 | DOI: 10.1038/ncomms14138 | www.nature.com/naturecommunications
